Error analysis on Regional and Global atmospheric transport model

In a situation where the detailed information of chemical transport is required, such as estimating the extent of dispersion and deposition of radioactive materials in Fukushima nuclear power plant accident, high-resolution regional climate models are often applied due to a computational burden of executing the corresponding global models. In the previous study, we looked at uncertainties specific to the regional model, such as a lateral boundary condition error, by comparing to the global model of the same model tendency with a shared dynamical core and physical schemes, and in this study, we will extend the idea to aerosol transport models and examine which physical processes of transport models are the most affected by the regional model bias.

A cloud-system resolving global climate model NICAM and its regional mode D-NICAM with the modified aerosol transport model SPRINTARS that emits and transports only the radioactive substance caesium-137 were compared. First, for validation of the modified SPRINTARS, a simulation follows the specifics of 2011 accidental releases of radioactive materials in Fukushima nuclear power plant. The regional D-NICAM is experimented with 2-km resolution centered at Fukushima, Japan, with an internal nudging of horizontal wind velocities, and compared to the ground-based observational data at several locations. Then, to quantify the effects of the bias in the transport model, we compared dispersion and deposition spread of caesium-137 between the global and regional simulations of the same model tendencies without internal nudging. Here, dry and wet depositions are separately examined and its discrepancies between regional and global simulations are analyzed with respect to the regional bias in velocities and precipitation. Thus we are trying to isolate and quantify the uncertainties of transport models originated from the uncertainties intrinsic to the regional models.